四步与八步图像生成器在步数蒸馏后的分支缓存
Branch Caching After Step Distillation in Four- and Eight-Step Image Generators
Mingyu Liang · Dingkun Xu
中文摘要
蒸馏后的图像生成器能否在剩余少量去噪步之间复用特征以节省时间?研究一种固定分支缓存,在保持学生模型噪声调度不变的前提下,替换选定的注意力与MLP计算。从同一初始噪声出发的一条完整轨迹和一条原始缓存轨迹,为每种分支类型和复用位置提供标量校正。在1024²分辨率下,一次内部复用将四步Qwen-Image-Lightning的去噪时间从1.279秒降至0.977秒。三次复用使八步Lightning获得1.51倍加速,八步Hyper-FLUX获得1.53倍加速。在全部200条DrawBench提示上评估图像质量。固定复用位置时,校正使两个Lightning学生模型的同种子图像保真度均值较原始复用分别提升0.66和0.87 dB。在Hyper-FLUX上PSNR差异为0.01 dB,而LPIPS从0.1625改善至0.1546。这些结果在保留各学生模型原始采样网格的情况下,衡量了分支缓存与校正的效果。
关键要点
- 01研究蒸馏后少步生成器能否在剩余去噪步间复用特征以节省时间
- 02固定分支缓存替换选定注意力与MLP计算,并用标量校正补偿误差
- 031024²下四步Qwen-Image-Lightning去噪从1.279秒降至0.977秒
- 04三次复用使八步Lightning加速1.51倍、八步Hyper-FLUX加速1.53倍
- 05200条DrawBench上校正提升同种子保真度,Hyper-FLUX的LPIPS从0.1625降至0.1546
解读
尚无解读。
原始英文摘要
arXiv:2611.25248v1 Announce Type: new Abstract: Can a distilled image generator save time by reusing features between its few remaining denoising steps? We study a fixed branch cache that replaces selected attention and MLP evaluations while preserving the student's noise schedule. One full trajectory and one raw-cache trajectory from the same initial noise supply a scalar correction for each branch type and reuse position. At $1024^2$, one internal reuse reduces four-step Qwen-Image-Lightning denoising time from 1.279 to 0.977 seconds. Three reuses yield $1.51\times$ speedup on eight-step Lightning and $1.53\times$ on eight-step Hyper-FLUX. We evaluate image quality on all 200 DrawBench prompts. With the reuse positions held fixed, correction improves mean same-seed image fidelity over raw reuse by 0.66 and 0.87 dB on the two Lightning students. On Hyper-FLUX the PSNR difference is 0.01 dB, while LPIPS improves from 0.1625 to 0.1546. These results measure the effect of branch caching and correction while retaining each student's original sampling grid.